John A Bittl1, Jacqueline E Tamis-Holland2, Christopher D Lang3, Yulei He4. 1. Munroe Regional Medical Center, Ocala, Florida. 2. Mount Sinai Saint Luke's Hospital and the Icahn School of Medicine, New York, New York. 3. Alpert School of Medicine, Brown University, Providence, Rhode Island. 4. Office of Research and Methodology, National Center for Health Statistics, Centers for Disease Control and Prevention, Hyattsville, Maryland.
Abstract
INTRODUCTION: During primary percutaneous coronary intervention (PCI), patients with ST-elevation myocardial infarction (STEMI) and multivessel coronary disease can undergo either multivessel intervention (MVI) or culprit-vessel intervention (CVI) only. BACKGROUND: Randomized controlled trials (RCTs) support the use of MVI, but cohort studies support the use of CVI. METHODS: We developed Bayesian models that incorporated parameters for study type and study outcome after MVI or CVI. RESULTS: A total of 18 studies (4 RCTs, 3 matched cohort studies, and 11 unmatched observational studies) enrolled 48,398 patients with STEMI and multivessel CAD and reported outcomes after MVI or CVI-only at the time of primary PCI. Using a Bayesian hierarchical model, we found that the point estimates replicated previously reported trends, but the wide Bayesian credible intervals (BCI) excluded any plausible mortality difference between MVI versus CVI in all three study types: RCTs (odds ratio [OR] 0.60, 95% BCI 0.31-1.20), matched cohort studies (OR 1.37, 95% BCI 0.86-2.24), or unmatched cohort studies (OR 1.16, 95% BCI 0.70-1.89). Both the global summary (OR 1.10, 95% BCI 0.74-1.51) and a sensitivity analysis that weighted the RCTs 1-5 times as much as observational studies revealed no credible advantage of one PCI strategy over the other (OR 1.05, 95% BCI 0.64-1.48). CONCLUSIONS: Bayesian approaches contextualize the comparison of different strategies by study type and suggest that neither MVI nor CVI emerges as a preferred strategy in an analysis that accounts mortality differences.
INTRODUCTION: During primary percutaneous coronary intervention (PCI), patients with ST-elevation myocardial infarction (STEMI) and multivessel coronary disease can undergo either multivessel intervention (MVI) or culprit-vessel intervention (CVI) only. BACKGROUND: Randomized controlled trials (RCTs) support the use of MVI, but cohort studies support the use of CVI. METHODS: We developed Bayesian models that incorporated parameters for study type and study outcome after MVI or CVI. RESULTS: A total of 18 studies (4 RCTs, 3 matched cohort studies, and 11 unmatched observational studies) enrolled 48,398 patients with STEMI and multivessel CAD and reported outcomes after MVI or CVI-only at the time of primary PCI. Using a Bayesian hierarchical model, we found that the point estimates replicated previously reported trends, but the wide Bayesian credible intervals (BCI) excluded any plausible mortality difference between MVI versus CVI in all three study types: RCTs (odds ratio [OR] 0.60, 95% BCI 0.31-1.20), matched cohort studies (OR 1.37, 95% BCI 0.86-2.24), or unmatched cohort studies (OR 1.16, 95% BCI 0.70-1.89). Both the global summary (OR 1.10, 95% BCI 0.74-1.51) and a sensitivity analysis that weighted the RCTs 1-5 times as much as observational studies revealed no credible advantage of one PCI strategy over the other (OR 1.05, 95% BCI 0.64-1.48). CONCLUSIONS: Bayesian approaches contextualize the comparison of different strategies by study type and suggest that neither MVI nor CVI emerges as a preferred strategy in an analysis that accounts mortality differences.
Authors: Héctor Cubero-Gallego; Rafael Romaguera; Albert Ariza-Sole; Joan Antoni Gómez-Hospital; Angel Cequier Journal: Cardiovasc Diagn Ther Date: 2017-06